Antonio Lieto

We introduce a framework called LAPITHS (Language-model Analysis through Paradigm-grounded Interpretations of Theses about Human-likenesS) and use it to show that several major claims advanced by models such as CENTAUR—proposed as an artificial Unified Model of Cognition—are not theoretically or empirically justified. LAPITHS provides a principled reference point for counteracting the current behaviouristic tendency in AI research to interpret the human-level performances of transformer-based language models as evidence of human-like underlying computation and, by extension, as signs of cognitive abilities. The novelty of LAPITHS lies in making explicit the arguments grounded in two quantitative assessments: (i) the Minimal Cognitive Grid, a theoretically motivated method for estimating the cognitive plausibility of artificial systems, and (ii) a behavioural comparison showing that results similar to those reported for CENTAUR-like models can be reproduced by other systems that do not satisfy the structural constraints typically associated with cognitive plausibility, and whose outputs do not provide independent explanatory insight into human cognition.

In this paper, we employ the Minimal Cognitive Grid (MCG), a framework created to evaluate the cognitive plausibility of artificial systems, to offer a systematic assessment of leading computational models of analogy and metaphor, including the Structure-Mapping Engine (SME), CogSketch, METCL, and Large Language Models (LLMs). We present a formal and quantitative operationalization of the MCG framework and, through the analysis of its three main dimensions (Functional/Structural Ratio, Generality, and Performance Match), examine how well each system aligns with standard cognitive theories of the modeled phenomena, thus allowing for comparison of the models with respect to their cognitive plausibility, according to consistent and generalizable mathematical criteria.

In this work we present and test a RAG-based model called RAGGAE (i.e. RAG for the General Analysis of Explanans) tested in the context of Human Explanation of Robotic BehaviorS (HERBS). The RAGGAE model makes use of an ontology of explanations, enriching the knowledge of state of the art general purpose Large Language Models like Google Gemini 2.0 Flash, DeepSeek R1 and GPT-4o. The results show that the combination of a general LLM with a symbolic, and philosophically grounded, ontology can be a useful instrument to improve the investigation, identification and the analysis of the types of explanations that humans use to verbalize - and make sense of - the behavior of robotic agents. Keywords: Human-Robot Interaction · Explanations · RAG · Large Language Models · Ontology.

We propose a system called METCL (Metaphor Elaboration in Typicality-Based Compositional Logic) able to generate and identify metaphors by using the TCL reasoning framework, specialized in human-like commonsense concept combination. We show that METCL is able to improve both state-of-the-art Large Language Models (e.g. DeepSeek-R1, GPT-4o, Qwen2.5-Max) and symbolic ones in the task of metaphor identification. Additionally, we show how the metaphors generated by METCL are generally well accepted by human subjects. The obtained results are encouraging and pave the way to research in automatic metaphor generation and comprehension based on the assumption that metaphors interpretation can be partially regarded as a categorization problem relying on generative commonsense concept combination.

I will present the rationale followed for the conceptualization and the following development the Dual PECCS system that relies on the cognitively grounded heterogeneous proxytypes representational hypothesis [Lieto 2014]. Such hypothesis allows integrating exemplars and prototype theories of categorization as well as theory-theory [Lieto 2019] and has provided useful insights in the context of cognitive modelling for what concerns the typicality effects in categorization [Lieto, 2021]. As argued in [Lieto et al., 2018b] a pivotal role in this respect is played by the use of the conceptual spaces framework and by its integration with a symbolic knowledge representation layer.

Persuasive technologies can adopt several strategies to change the attitudes and behaviors of their users. In this work I synthesize the lessons learned from three empirical case studies on automated persuasion that have been carried out in the last decade in the contexts of: persuasive news recommendations, social robotics, and e-commerce, respectively. In particular, such studies have assessed, in the technological domain, the effects of nudging techniques relying on well known persuasive argumentation schemas and on framing strategies. In discussing the main findings, I will argue that the obtained persuasive effects are due to the fact that such techniques leverage on cognitive mechanisms that refer to the “system 1” types of automatic processes hypothesized in the context of the dual process theory of reasoning. As a consequence of this state of affairs, any automated persuasive systems (used for ethical purposes: e.g. suggesting good health habits) should be able to re-use these types of system 1 strategies in order to gradually take the users in an argumentation territory where system 2 processes can take place.

Inventing novel knowledge to solve problems is a crucial, creative, mechanism employed by humans, to extend their range of action. In this paper, we present TCL (typicality-based compositional logic): a probabilistic, non monotonic extension of standard Description Logics of typicality, and will show how this framework is able to endow artificial systems of a human-like, commonsense based, concept composition procedure that allows its employment in a number of applications (ranging from computational creativity to goal-based reasoning to recommender systems and affective computing). The framework relies on 3 main ingredients: a non monotonic extension of standard Description Logics (allowing to reason on exceptions to inheritance in standard knowledge bases) a probabilistic extension coming from the field of logic programming (the DISPONTE semantics) a cognitive heuristics known ad HEAD-Modifier determining preference rules for the inheritance mechanisms of the novel concepts to be generated via commonsense rules. We report the obtained results and the lessons learned of this research path in the context of model-based AI applications.

In this work we present NERVOUS, an intelligent recommender system exploiting a probabilistic extension of a Description Logic of typicality to dynamically generate novel contents in AllMusic, a comprehensive and in-depth resource about music, providing data about albums, bands, musicians and songs. The tool can be used for both the generation of novel music genres and styles, described by a set of typical properties characterizing them, and the reclassification of the available songs within such new genres.

Commonsense reasoning is one of the main open problems in the field of Artificial Intelligence (AI) while, on the other hand, seems to be a very intuitive and default reasoning mode in humans and other animals. In this lecture, I will present the TCL reasoning framework that has been developed to address  the problem of dynamic, goal-directed, knowledge invention and will show how it has been applied to different case studies and applications in the areas of cognitive robotics, cognitive architectures and computational creativity. 

In this work we present an explainable system for emotion attribution and recommendation (called DEGARI (Dynamic Emotion Generator And ReclassIfier) relying on a recently introduced probabilistic commonsense reasoning framework (i.e. the TCL logic, see Lieto & Pozzato 2020) which is based on a human-like procedure for the automatic generation of novel concepts in a Description Logics knowledge base (see also Lieto et al. 2019, Chiodino et al. 2020 for other applications). In particular, in order to model human-like forms of concept combinations, the TCL logic combines a probabilistic description logics of typicality with the HEAD-MODIFIER heuristics coming from cognitive semantics. In the context of our application of such framework, our system exploits the logic TCL to automatically generate novel commonsense semantic representations of compound emotions (e.g. Love as derived from the combination of Joy and Trust according to Plutchik’s theory of emotions, see Plutchik 2001). The generated emotions corresponds to prototypes, i.e. commonsense representations of given concepts, and have been used to reclassify emotion-related contents in a variety of artistic domains. We have tested our system in the context of the H2020 EU project SPICE for providing artistic recommendations of museum items in the Galleria di Arte Moderna (GAM) of Turin. The obtained results (reported in Lieto et al, 2021 and Lieto at al., to appear) show promising results for what concerns both the acceptance of the provided affective recommendation and the explainability associated to each suggestion made by the users. Here we discuss, the open problems and the lessons learned.

In this article, I argue that the artificial components of hybrid bionic systems do not play a direct explanatory role, i.e., in simulative terms, in the overall context of the systems in which they are embedded in. More precisely, I claim that the internal procedures determining the output of such artificial devices, replacing biological tissues and connected to other biological tissues, cannot be used to directly explain the corresponding mechanisms of the biological component(s) they substitute (and therefore cannot be used to explain the local mechanisms determining an overall biological or cognitive function replicated by such bionic models). I ground this analysis on the use of the Minimal Cognitive Grid (MCG), a novel framework proposed in Lieto (Cognitive design for artificial minds, 2021) to rank the epistemological and explanatory status of biologically and cognitively inspired artificial systems. Despite the lack of such a direct mechanistic explanation from the artificial component, however, I also argue that the hybrid bionic systems can have an indirect explanatory role similar to the one played by some AI systems built by using an overall structural design approach (but including the partial adoption of functional components). In particular, the artificial replacement of part(s) of a biological system can provide i) a local functional account of that part(s) in the context of the overall functioning of the hybrid biological–artificial system and ii) global insights about the structural mechanisms of the biological elements connected to such artificial devices.

The mental rotation ability is an essential spatial reasoning skill in human cognition and has proven to be an essential predictor of mathematical and STEM skills, critical and computational thinking. Despite its importance, little is known about when and how mental rotation processes are activated in games explicitly targeting spatial reasoning tasks. In particular, the relationship between spatial abilities and TetrisTM has been analysed several times in the literature. However, these analyses have shown contrasting results between the effectiveness of Tetris-based training activities to improve mental rotation skills. In this work, we studied whether, and under what conditions, such ability is used in the TetrisTM game by explicitly modelling mental rotation via an ACT-R based cognitive model controlling a virtual agent. The obtained results show meaningful insights into the activation of mental rotation during game dynamics. The study suggests the necessity to adapt game dynamics in order to force the activation of this process and, therefore, can be of inspiration to design learning activities based on TetrisTM or re-design the game itself to improve its educational effectiveness.

In the last decades Human-Computer Interaction (HCI) has started to focus attention on “persuasive technologies” having the goal of changing users’ behavior and attitudes according to a predefined direction. In this talk we show how some of the techniques employed in such technologies trigger some well known cognitive biases by adopting a strategy relying on logical fallacies (i.e. forms of reasoning which are logically invalid but psychologically persuasive). In particular, we will show how the mechanisms reducible to logical fallacies are used to design web and mobile interfaces in domains ranging from the e-commerce to the jihadist propaganda. The final part of the talk will be devoted to point out the potential ethical dangers related to the misuse of these techniques in the design of persuasive technologies.

I will review the main problems concerning commonsense reasoning in machines and I will present resent two different applications - namaly: the Dual PECCS linguistic categorization system and the TCL reasoning framework that have been developed to address, respectively, the problem of typicality effects and the one of commonsense compositionality, in a way that is integrated or compliant with different cognitive architectures thus extending their knowledge processing capabilities In doing so I will show how such aspects are better dealt with at different levels of representation and will discuss how the adoption of a cognitively inspired approach be useful in the design and implementation of the next generation AI systems mastering commonsense.

I will present the rationale followed for the conceptualization and the following development the Dual PECCS system that relies on the cognitively grounded heterogeneous proxytypes representational hypothesis. Such hypothesis allows integrating exemplars and prototype theories of categorization and has provided useful insights in the context of cognitive modelling for what concerns the typicality effects in categorization. As argued in [Chella et al., 2017] [Lieto et al., 2018b] [Lieto et al., 2018a] a pivotal role in this respect is played by the use of the conceptual spaces framework and by its integration with a symbolic knowledge representation layer.

I will present two different applications - Dual PECCS and the TCL reasoning framework - addressing some crucial aspects of commonsense reasoning (namely: dealing with typicality effects and with the problem of commonsense compositionality) in a way that is integrated or compliant with different cognitive architectures. In doing so I will show how such aspects are better dealt with at different levels of representation and will discuss the adopted solution to integrate such representational layers

A 3rd person Knowledge Level analysis of cognitive architectures Abstract I provide a knowledge level analysis of the main representational and reasoning problems affecting the cognitive architectures for what concerns this issue. In providing this analysis I will show, by considering some of the main cognitive architectures currently available (e.g. SOAR, ACT-R, CLARION), how one of the main problems of such architectures is represented by the fact that their knowledge representation and processing mechanisms are not sufficiently constrained with “structural insights” (Lieto 2021) coming from cognitive science for dealing with commonsense knowledge and reasoning (Lebiere, Oltramari, 2018). As a possible way out to such knowledge processing issues, I present the main assumptions that have led to the development of the Dual PECCS categorization system (Lieto, Radicioni, Rho 2017) and discuss some of the lessons learned and their possible implications in the design of the knowledge modules and knowledge-processing mechanisms of integrated cognitive architectures.

The paper presents the heterogeneous proxytypes hypothesis as a cognitively-inspired computational framework able to reconcile, in both natural and artificial systems, different theories of typicality about conceptual representation and reasoning that have been traditionally seen as incompatible. In particular, through the Dual PECCS system and its evolution, it shows how prototypes, exemplars and theory-theory like conceptual representations can be integrated in a cognitive artificial agent (thus extending its categorization capabilities) and, in addition, can provide useful insights in the context of a computationally grounded science of the mind.

Commonsense reasoning is one of the main open problems in the field of Artificial Intelligence (AI) while, on the other hand, seems to be a very intuitive and default reasoning mode in humans and other animals. In this talk, we discuss the different paradigms that have been developed in AI and Computational Cognitive Science to deal with this problem (ranging from logic-based methods, to diagrammatic-based ones). In particular, we discuss - via two different case studies concerning commonsense categorization and knowledge invention tasks - how cognitively inspired heuristics can help (both in terms of efficiency and efficacy) in the realization of intelligent artificial systems able to reason in a human-like fashion, with results comparable to human-level performances.

Commonsense reasoning is a crucial human ability employed in everyday tasks. In this talk I provide a knowledge level analysis of the main representational and reasoning problems affecting the cognitive architectures for what concerns this issue. In providing this analysis I will show, by considering some of the main cognitive architectures currently available (e.g. SOAR, ACT-R, CLARION), how one of the main problems of such architectures is represented by the fact that their knowledge representation and processing mechanisms are not sufficiently constrained with insights coming from cognitive science (Lieto 2021; Lieto, Lebiere, Oltramari, 2018). As a possible way out to such knowledge processing issues, I present the main assumptions that have led to the development of the Dual PECCS categorization system (Lieto, Radicioni, Rho 2017) and discuss some of the lessons learned and their possible implications in the design of the knowledge modules and knowledge-processing mechanisms of integrated cognitive architectures.

Book Description (Blurb): Cognitive Design for Artificial Minds explains the crucial role that human cognition research plays in the design and realization of artificial intelligence systems, illustrating the steps necessary for the design of artificial models of cognition. It bridges the gap between the theoretical, experimental and technological issues addressed in the context of AI of cognitive inspiration and computational cognitive science. Beginning with an overview of the historical, methodological and technical issues in the field of Cognitively-Inspired Artificial Intelligence, Lieto illustrates how the cognitive design approach has an important role to play in the development of intelligent AI technologies and plausible computational models of cognition. Introducing a unique perspective that draws upon Cybernetics and early AI principles, Lieto emphasizes the need for an equivalence between cognitive processes and implemented AI procedures, in order to realise biologically and cognitively inspired artificial minds. He also introduces the Minimal Cognitive Grid, a pragmatic method to rank the different degrees of biologically and cognitive accuracy of artificial systems in order project and predict their explanatory power with respect to the natural systems taken as source of inspiration. Providing a comprehensive overview of cognitive design principles in constructing artificial minds, this text will be essential reading for students and researchers of artificial intelligence and cognitive science.

Inventing novel knowledge to solve problems is a crucial, creative, mechanism employed by humans, to extend their range of action. In this talk, I will show how commonsense reasoning plays a crucial role in this respect. In particular, I will present a cognitively inspired reasoning framework for knowledge invention and creative problem solving exploiting TCL: a non-monotonic extension of a Description Logic (DL) of typicality able to combine prototypical (commonsense) descriptions of concepts in a human-like fashion. The proposed approach has been tested both in the task of goal-driven concept invention and has additionally applied within the context of serendipity-based recommendation systems. I will present the obtained results, the lessons learned, and the road ahead of this research path.

This paper presents a practical case study showing how, despite the nowadays limited collaboration between AI and Cognitive Science (CogSci), cognitive research can still have an important role in the development of novel AI technologies. After a brief historical introduction about the reasons of the divorce between AI and CogSci research agendas (happened in the mid’80s of the last century), we try to provide evidence of a renewed collaboration by showing a recent case study on a commonsense reasoning system, built by using insights from cognitive semantics.